JP4991431B2 - External environment control device based on brain function measurement - Google Patents

Description

  The present invention relates to an external environment control device using brain function measurement, and is particularly suitable for a biological control device that uses an output signal of an information input device using an optical biological measurement method as an input signal of an external device that performs various controls. is there.

  The external environment control device based on brain function measurement called a biological control device uses optical biometrics and a biological input device to control various devices that do not use buttons, mice, or handles, and to perform body movements such as falling asleep or reducing attention. Triggering alarm devices based on conditions, control according to the consciousness / unconscious state of devices in the surrounding environment, judgment of learning effects and rehabilitation proficiency, display of senses and thoughts of infants, sick people, animals, etc. Applied to.

  Conventionally, various input devices such as buttons, a keyboard, and a mouse have been used to operate devices such as information processing devices, games, home appliances, AV devices, and transport devices. However, such an input device that is operated by a human limb may temporarily prevent a user from concentrating on another input or may reduce the sense of realism in play. In addition, difficulties may arise when a disabled person or the like operates. Thus, many attempts have been made to control devices using information obtained directly from the brain.

  As means for directly obtaining the control information as described above by measuring the brain, research has been conducted on techniques for directly picking up electrical signals by inserting electrodes into the brain. However, there is a risk of side effects in such a measurement (invasive method) where it is necessary to insert a scalpel into the human body and implant an implant device. On the other hand, positron emission tomography (PET), functional magnetic resonance imaging device (functional magnetic resonance imaging: fMRI) as a measurement method (non-invasive method) that does not use a means to open the human body directly such as a needle or scalpel. And electroencephalography.

  Patent Document 1 discloses a device that directly inputs control information from the brain to a controlled device using an electroencephalogram. In the apparatus disclosed in Patent Document 1, a computer, particularly a game machine, is to be controlled by inputting an electroencephalogram as it is to an information processing apparatus as in the case of measuring an electrocardiogram. Such a direct input device from the brain to the controlled device is capable of controlling an external device of a patient whose motor function is impaired, and is expected to contribute to the social participation of physically disabled people.

  In addition to these technologies, using near-infrared spectroscopy, blood volume changes in the cerebral cortex associated with brain activity are measured at multiple points, and changes in the blood volume are displayed as images (biological light measurement). Device) has already been put into practical use and is disclosed in Non-Patent Document 2, for example.

Examples of disclosure of background art in this field include Patent Documents 3 to 6 and Non-Patent Documents 1 to 3 in addition to Patent Documents 1 and 2 described above. The contents of disclosure of these documents will be described as needed in the section of the present invention, solutions, or embodiments.
Japanese Patent Laid-Open No. 7-124331 Japanese Patent No. 3543453 Japanese Patent Laid-Open No. 9-1498949 JP 2000-172407 A JP 2002-119511 A JP-A-2005-13464, biological light measurement device. Maki, A et al., (1995) “Spatial and temporal analysis of human motor activity using noninvasive NIR topography” Medical Physics 22, 1997-20 HYPERLINK "http://www.hitachi.co.jp/New/cnews/month/2005/09/0926.html" http://www.hitachi.co.jp/New/cnews/month/2005/09/ 0926.html, “Commercialized Yes / No Judgment Device for ALS Patients Who Cannot Move at All” VN Vapnik, "The Nature of Statistical Learning Theory, 2nd Ed.", Springer 2000.

  Hereinafter, in order to clarify the problem of the present invention, a basic technique of an external environment control device based on brain function measurement will be described with reference to FIGS. 1 to 5. 1 to 5 also show a basic configuration for explaining an embodiment of the present invention to be described later. FIG. 1 is a schematic diagram for explaining a device configuration example of a measurement method in an external environment control device based on brain function measurement, and shows a device configuration of a measurement method disclosed in Patent Literature 2, Non-Patent Literature 1, and the like. is there. Further, FIG. 2 shows a light path 201 propagating between a holder 107 for fixing the optical fiber 104 connected to the light irradiator 103 and a holder 107 for fixing the optical fiber 106 connected to the photodetector 105. It is a figure explaining.

  In FIG. 1, a subject 101 wears a helmet (probe) 102 for measurement. The helmet 102 includes an optical fiber 104 connected to a light irradiator 103 typified by a light emitting diode, a semiconductor laser, and a lamp, an optical fiber 106 connected to a photodetector 105 typified by an avalanche photodiode and a photomultiplier tube. An optical fiber holder 107 for connection is provided.

  The optical fiber tip 108 shown in the optical fibers 104 and 106 scrapes the hair of the subject 101 and makes light contact with the scalp. This is because the optical transmission efficiency decreases when the hair is sandwiched between the tip of the optical fiber and the scalp. A plurality of the light irradiators 103 are provided, and the output light intensity for each time constitutes a multi-channel system managed by the control device 109. The control content is transmitted to the signal processing device 111 connected to the photodetector 105 using the transmission cable 110, and may be used for estimating the intensity change of the light that has passed through the living body by the signal processing device 111. is there. Reference numeral 112 is an information processing apparatus that constitutes an input unit typified by a personal computer and a workstation, transmits the control content to the control apparatus 111 using the transmission cable 113, and takes in the processing result, Perform analysis processing. The analysis result is displayed on the display screen 114. Reference numerals 115 and 116 denote a keyboard and a mouse which are input devices of the information processing apparatus 112.

  The holders 107A and 107B are fixed to a helmet 102 that fits the head of the subject 101 shown in FIG. 1, and the tips of the optical fibers 104 and 106 are in contact with the scalp of the object to be examined. FIG. 2 shows a typical human brain structure. This brain structure is composed of a scalp 202, a skull 203, a cerebrospinal fluid layer 204, a cerebral cortex 205, and the like. These biological tissues have optical scattering characteristics and absorption characteristics, and it is known that the light scattering characteristics of the skull 203 are particularly large. For this reason, it is known that the light emitted from the light irradiator 103 is scattered by the light scattering characteristics, and the intensity is gradually lost due to the light absorption characteristics. Here, the holders 107 shown in the figure are arranged in a grid pattern at intervals of about 30 mm.

  When arranged at such an arrangement interval, the light emitted from the optical fiber 104 connected to the light irradiator 103 propagates through the living tissue following a path shown as a shape 210 in the figure while repeating scattering and absorption. The light reaches the optical fiber 106 connected to the photodetector 105 and is detected. For this measurement, near-infrared light (wavelength: 600 nm to 900 nm) having high permeability to living tissue is used. Reference numeral 211 in the figure indicates a region where the concentration of in vivo metabolites represented by blood volume has increased with brain activity in the cerebral cortex 205.

  Blood is composed of various substances, among which hemoglobin (oxygenated hemoglobin and deoxygenated hemoglobin are known to absorb near-infrared light used for measurement. It is known that the intensity of the detected light is attenuated as it increases, that is, the change in blood volume can be estimated by detecting the change in the detected light. Non-Patent Document 1.

  A major feature of the biological light measurement apparatus described in Non-Patent Document 1 is that such a light irradiator and a light detector are two-dimensionally arranged on the scalp of the subject. As a result, it is possible to visualize the distribution of blood volume changes accompanying brain activity.

  FIG. 3 shows an arrangement position 301 (indicated by white circles) of the optical fiber holder 107A connected to the light irradiator 103 and an arrangement position 302 (shown in the figure) of the optical fiber holder 107B connected to the photodetector 105. (Black circle) of FIG. These irradiation / detection holders are spatially arranged alternately at intervals of about 30 mm. According to FIG. 2, the thickness of the shape 210 of the optical path is determined by the arrangement position 301 of the optical fiber holder 107 </ b> A connected to the light irradiator 103 and the arrangement position 302 of the optical fiber holder 107 </ b> B connected to the photodetector 105. It is the maximum at a position 303 immediately below the midpoint. For this reason, it is known that the sensitivity to changes in blood volume is maximized at this midpoint. Therefore, this midpoint (indicated by an X in the figure) is called a sampling point 303, which is a point that gives positional information of blood volume changes detected by a pair of optical fibers.

  In FIG. 3, the arrangement position 301 of the optical fiber holder 107A connected to the light irradiator 13, the arrangement position 302 of the optical fiber holder 107B connected to the photodetector 105, and the sampling point 303 are represented by two representative arrangement positions and one Only the sampling points are provided with reference numerals. As can be seen from the fact that the positions and sampling points of the other holders are also indicated by the same white circles or black circles and the display of the X symbol, in the example of FIG. 3, 24 inputs are provided by 8 inputs and 8 detectors. Sampling points exist. Using the information of the 24 sampling points, it is possible to visualize the brain activity as illustrated in FIG.

  FIG. 4 shows an example of a topography image 401 obtained by spatially interpolating changes in blood volume at each sampling point. With this image, it is possible to obtain a spatial distribution of changes in blood volume at a certain time, and it is also possible to image an average value of changes in blood volume during the brain activity period. As shown in FIG. 4, it is possible to measure brain activity by using the biological light measurement device. Further, as shown in the figure, this image can be displayed by superimposing the distribution of the arrangement position 301 of the irradiation optical fiber holder 107A, the arrangement position 302 of the detection optical fiber holder 107B, and the sampling points 303. This makes it possible to estimate the position of localized brain activity.

  In this measurement method, measurement can be performed simply by wearing a helmet using weak light. Therefore, it is possible to perform inspection with high safety and a free posture. For this reason, it is possible to measure a wide range of people from infants to elderly people. For example, in magnetic resonance imaging devices and positron emission tomography that have been used for measurement in the past, it is not allowed to move inside the measurement, especially in the case of infants, in order to suppress movement, anesthesia and sedatives Was sometimes administered. In such a case, it was difficult to accurately measure brain function activation. In addition, regardless of children or the elderly, the subject is generally inspected in a closed space and cannot move, so the problem is that it is mentally different from normal living conditions. Although the measurement of brain waves can be taken relatively stably, it is strongly affected by electromagnetic waves, so there are cases where an electromagnetic shield is required for stable operation. In contrast, a brain function measurement device that uses light can be tested in a free environment and in a free posture, so it can be used in situations where measurement was difficult with conventional brain function measurement devices. Become.

Patent Documents 3 and 4 propose examples in which an external device is controlled using a technique for measuring the amount of blood volume fluctuation in the cerebral cortex using infrared rays. This patent uses the knowledge that the human brain is divided into different cell structures, as represented by the Broadman brain map, and that each region shares different functions. . For example, when looking at the brain from the side, the area related to spontaneous movement (hands, fingers, feet, etc.) is the top, the area related to sensation, vision, etc. is the back of the head, and the area related to language is the left half. Each division is responsible.
An intention communication device for patients whose muscles cannot be moved due to a neurological disease using the function of such a measuring device has been announced (Non-patent Documents 3 and 5). By applying near infrared spectroscopy to two parts of the forehead, it measures the change in cerebral blood volume when performing calculations, singing, etc. in the head. .

  There is a method called a support vector machine as a method used for deriving condition judgment from data of a plurality of channels. Based on the contents described in Non-Patent Document 4, an outline of the method will be described. It is assumed that there are N pieces of learning data (x, y) expressed by a combination of a vector x having d-dimensional information and a discrete value y indicating values {+1, −1} giving the learning rule. FIG. 5 shows a schematic diagram thereof. Data 501 shown as a white circle symbol in FIG. 5 represents a learning state 1 shown as y = 1, and data 502 shown as a black circle symbol represents a learning state 2 shown as y = -1.

At this time, in order to separate the learning data 1 and 2, the D-1 dimensional hyperplane 503 (expressed by Equation 1) in FIG. 5 is defined so as to satisfy the optimization condition shown in Equation 2.
This is equivalent to defining the hyperplane so that the distances from the two groups are equal. In addition, when it is impossible to separate on a D-1 dimensional plane, it is defined as follows using a positive slug variable (soft margin). Also, a method is known in which a new Hilbert space is created by a conversion formula and the support vector machine is optimized in that space (kernel trick).

  As disclosed in Patent Documents 1 and 2, the number of types of brain activity that can be measured increases by installing more sensors in the helmet. For this reason, there is a possibility that the degree of freedom of device control is increased. However, when many sensors are installed in a helmet, all sensors do not always function normally with a single installation. For example, when the hair is caught in the contact tip portion 108 of the optical cable, the light spectrum is significantly absorbed at that location, and the accuracy of the detector may be significantly deteriorated. In order to avoid this, Patent Document 5 proposes a method of wearing a helmet while checking the light intensity of each sensor.

  In this way, one of the most important issues when considering the biological optical measurement device that is expected to be effectively used in various applications is to drive external devices and use in communication. The stability of the analytical means. By allowing the helmet to be easily worn and measuring in a short time, the apparatus does not place a burden on the user.

  However, when attempting to control equipment using this signal, the human brain structure and the response signal obtained from it are subject to individual differences and uncertainties in sensor installation. This is a major issue. Examples of this individual difference include skin thickness, hair arrangement, physical size and shape of the skull, geometric errors in the boundaries of the brain's functionally localized areas, and differences in blood vessel structure There are differences in response due to fluctuations in blood volume. Moreover, even if the same task of the same calculation is given in the same way, there are people who visually solve the solution and others who seek the solution by phoneme. For this reason, there are not a few cases where it is difficult to predetermine a reaction pattern that always adapts to any human being as a generalized theory.

  And depending on the installation state of the helmet 12 shown in FIG. 1, there exists a case where a part of measurement location cannot measure well. As described above, this is conspicuous when the holders 17A and 17B are located in a place where there is a lot of hair and the contact portion of the optical fiber is attached. Further, as the helmet 12 is remounted, the signal system may change due to a slightly different position of the sensor of the light source to be installed or the degree to which the helmet 12 presses the scalp. . Furthermore, there is a case where the influence amount of the pulse with respect to the change in the blood volume of the brain varies depending on the posture of the user.

  In addition, since the conditions such as the thickness of the skull and the pigment state of the skin are different at each measurement point, the degree of light absorption in the areas such as the epidermis and the bone varies from person to person and from place to place. For this reason, near-infrared spectroscopy cannot directly compare data observed at each measurement point as an absolute value. In the operation of optical measurement, in a state where a measurement device is actually installed, a certain moment at the start of measurement is set as a zero point, and a relative variation value from that state is handled as information of a measurement result. In connection with this problem, the current technology that cannot completely measure and estimate the optical path that passes through the human body, it is necessary to stably determine the installation state of the helmet based only on the absolute value of the measurement contents. There will be difficulties.

  For these reasons, factors corresponding to the measurement accuracy and error of measurement vary from individual to individual and from measurement to measurement. For example, an element that a reaction is likely to occur for a reason based on the functional localization of the brain at a certain measurement point may be different from an element that has a high reproducibility of accuracy for detecting the reaction. In other words, it shows a strong response when it can be measured correctly, but it is often difficult to measure because the installation state of the optical input / output device (the optical fiber holder 107 in FIG. 1) tends to deteriorate, and the measured response. There are many situations in which the stability of repeated measurement shows fluctuations due to different reasons, such as measurement locations where the installation state of the device 17 is always stable. Since there are variations in reliability that differ from one measurement location to another, it is difficult to maintain the stability of condition determination simply by applying an existing method such as a support vector machine simply and uniformly.

  An object of the present invention is to realize an external environment control device using brain function measurement with improved reliability by improving measurement stability.

  In the present invention, information collected from each sensor is divided into individual information processing modules and processed independently, and each reflects information reflecting functional localization and pulse information independently. Analyze, classify, and manage the degree of reliability accuracy repeatability for each helmet setting and the intensity of signal detection for response to the subject's introspection activities for each element of the information processing module. Save additionally. This avoids the failure of one sensor from spreading to the entire system, and the module contributes to the final information output in accordance with the reliability accuracy derived from the check result of whether each module is functioning. Change weights or give instructions to users to change tasks.

  According to the present invention, even when an appropriate signal is not obtained compared with the past history, it is possible to separate whether the problem is due to a biological reaction problem associated with the execution of the task or based on the contact of the device. It becomes possible to select a part to be corrected and an appropriate reflection task to be performed as an instruction under the environment. In addition, it is possible to create an interface environment that leads to more stable behavior by holding and growing learning parameters based on the individual's execution history for each component of the functional element for a long period of time, and it can be transferred to another measurement environment. It becomes like this.

  Embodiments of an external control device based on biological light measurement according to the present invention will be described below in detail with reference to the drawings.

  FIG. 6 is a schematic diagram showing Example 1 of an external control device based on biological light measurement according to the present invention. The same functions as those shown in FIG. 1 are denoted by the same reference numerals. The calculation processing mechanism system according to the first embodiment includes information processing devices such as an input unit 112, an information processing unit 601, and an output unit 602. It is assumed that the information processing devices are connected to each other on a network and can exchange input / output results. As information processing equipment of this system, general-purpose computing equipment such as a well-known PC (personal computer, personal computer) or built-in microcomputer module can be used.

  Further, the processing of the input unit 112, the information processing unit 601, and the output unit (output module) 602 can be implemented as a software module in the same information processing device. In order to facilitate this, it is shown in the form of an individual information processing device. Further, a screen monitor 603 for performing the result of the output unit (output module) 602 on the subject or presenting a request from the information processing unit 601 to the subject is disposed in front of the subject 101. A medium 605 for recording the contents of the measurement history of the subject 101 and a reader 604 for transmitting the contents to the information processing unit 601 are arranged.

  In addition, information on how to combine modules used in the information processing unit 601 is shown on the medium 608, and this information is read from the reader 607 when the information processing device 601 is activated. This information can also be changed by the manual input interface 606.

  The apparatus according to the first embodiment operates by switching between two modes of a learning phase mode and a real-time operation phase mode. The operation procedure of the entire apparatus of the present embodiment will be described using the flowchart of FIG. In the measurement phase mode, this apparatus measures the state of cerebral blood volume when the subject is performing a given task and generates parameters for use in the real-time operation mode. In the real-time operation mode, using the measured cerebral blood volume state and the parameters created in the learning phase mode, the task being performed by the subject is inferred and the operation to the environment presented to the subject is executed. To do. The subject switches the task while grasping the state of the environmental operation, and performs the intended action. Even in the real-time operation mode, the parameters are readjusted according to the performance level of this action execution. The following describes the behavior of the device in each mode.

  The operation of the first embodiment is started (step 701), and the user 101 installs the helmet 107 of the optical measuring device as a sensor on the head (step 702). Thereafter, the operation state of each agent is confirmed in the learning phase mode (step 703), and the amount of agent reaction is estimated from the degree of achievement of the current response (step 704). In accordance with the analogy result, the user is advised to change the use task or to re-install the optical fiber holder (also referred to as a probe holder) 107 (step 705). The user 101 re-installs the optical probe holder 107 in accordance with these presentations or determines a task to be used, and moves to the real-time operation phase mode (step 706). The specific contents of each process are explained below.

  In the information processing apparatus 601, an algorithm part for converting a blood volume variation measured in each part of the brain sent as a signal from the optical probe holder 107 into an input value of the output module 602 is input as a program. Saved. There are a feedback operation program in which parameter adjustment processing is performed in the measurement phase mode, and a program used in the output module 602 in the real-time operation phase mode, and the calculation module is shared and used in these two modes. First, the contents of the calculation processing algorithm performed in the information processing unit 601 will be described.

FIG. 8 is a system configuration diagram illustrating the structure of a module that performs information processing, and schematically illustrates the modules that configure the information processing process described above. This system
It consists of three components. The module 801 receives measurement data at each measurement point from information sent from the input unit 112. Also, it has a function of holding the past value. Hereinafter, this module 801 is referred to as a “sampler”. The parts for filtering the information stored in the sampler are analysis parts 803 and 804.

Each module (802, 803, 804, 805) described as “Agent” in FIG. 8 is mounted on the program as various filter components to be described later, and is hereinafter referred to as “agent”. Further, the integration part 805 integrates the output information of each agent by weighted linear sum and transmits it to the external output 602 or uses it for calculation of parameter adjustment. Hereinafter, this part is called a “synthesizer”. The overall connection structure is a tree structure like a multi-level neural network, and the input and output of each agent module is time-series data, filtering functions specialized for biological measurement of cerebral blood volume, and general-purpose It consists of the filter calculation process.

  First, an agent module that performs filter processing will be described. The agents 802, 803, 804, and 805 are unit modules that receive time-series information that changes in real time, convert the information using a certain algorithm, and output the converted time-series information. These agents 802, 803, 804, and 805 have time series information updated in real time as input, and always output the processed results. These modules can have built-in constants, and the values are changed / adjusted by feedback from the synthesizer part or manual input from the manual input interface 606 to perform optimization processing.

  Agents are managed by being classified into modules of a primary agent 802, a secondary agent 803, a tertiary agent 804, and a pseudo agent 805 according to functions. A module that directly processes and outputs time-series information of blood volume fluctuation at a certain measurement point is called a primary agent. The primary agent 802 calculates the blood volume of oxygenated hemoglobin, the blood volume of deoxygenated hemoglobin, the total blood volume of hemoglobin, and the value obtained by subtracting the blood volume of deoxygenated hemoglobin from the blood volume of oxygenated hemoglobin. Can be created.

  A module that takes in results output by a plurality of primary agents and outputs processing results based on a given algorithm is referred to as a secondary agent 803. The secondary agent receives and processes the output from a plurality of other agents as its own input.

  In addition, a module that converts learning results obtained from the synthesizer into non-linear conversion of the results of the primary and secondary agents so that the reaction corresponding to changes in the user's introspection state can be easily separated. This is called a tertiary agent 804.

  In addition to this, a module that does not reflect the data from the cerebral blood volume directly stored in the sampler but can be used as an agent module in a pseudo manner in calculation is referred to as a pseudo agent module 805.

  Examples of algorithms used as primary agent modules are listed below.

  Frequency cut agent: This is a module that extracts and outputs only information about the target frequency component by cutting high frequency components and low frequency components.

  Moving average agent: A module that outputs the difference between the long-period moving average and the short-period moving average.

  Exponential Moving Average Agent: A module that outputs the result of taking the difference between the long-term exponential moving average and the short-period exponential moving average.

  In addition, a pseudo agent 805 that is driven by external information input in synchronization with blood volume information of the brain reaction is installed. This is a module that is processed in the same procedure as the primary module, except for the response of cerebral blood volume. The values output by these modules can be used as inputs to modules subsequent to the secondary as with the primary agent. Examples of this are listed below.

  Stimulus presentation agent: An agent that behaves as an output in an external stimulus presentation device, which becomes 1 from the task start time and becomes 0 at the end of the task.

  External pulse agent: An agent that collects pulse information from locations other than the brain and reflects the results.

  Motion acceleration agent: An agent that outputs the absolute value of the acceleration measuring instrument attached to the user's body.

  The secondary agent module receives and converts the output of the primary agent or pseudo agent as input. Examples of algorithms that can be used for this secondary agent module are listed below.

  Difference agent: A module that compares the output of a certain primary agent with the value output by that agent a certain time ago and outputs the difference.

  Left-right difference agent: This is a module that takes the difference between the output of the primary agent at a certain measurement position and the output of the agent at the measurement position located opposite to the brain as an output.

  Average difference agent: This is a module that takes the difference between the output of the primary agent at a certain measurement position and the average of the output of the agent at all measurement positions and outputs it.

  Divergent agent: This is a module that takes the difference between the output of the primary agent at a certain measurement position and the average of the output of the agent at the measurement positions arranged around it, and outputs it.

  Frequency analysis agent: A module that outputs, as an output, a time series obtained by extracting the absolute amount of a specific component from a measured agent output by frequency analysis.

  Pulse group component agent: This is a module that extracts a component that appears to be a pulse group by frequency analysis from the output of the measured agent, and outputs the time series.

  Examples of algorithms that can be used for the tertiary agent module are described below.

Functional localization agent: For each functional localization part of the brain, collect agents corresponding to the channels that cover the functional range, and provide multidimensional information such as support vector machines for reaction information obtained by executing the tasks described later. It is an agent that performs weighting using a processing mechanism and extracts its features. Define a measurement area as a range that covers the N field in Broadman's brain map, use another brain function mapping formulation such as the 10/20 method used in EEG measurement, or estimate from the area For example, an agent that covers a language field in 90% of a standard head structure is defined according to a function to be performed.
When specifying the target area of the agent according to the function, there are individual differences such as the brain function and head shape varies for each subject, and the individual head size is different. Select a channel that covers the whole area. When the measurement is performed with a helmet placed on a subject, it is assumed that the language area function region is included by the region R1. However, it is assumed that it is a set of measurement points that are in the vicinity of an average language region estimated by the region R1, 10-20 method and show strong association with language-related tasks. Further, when the probe 17 is installed in another subject, the language area function region is assumed to be included by the region R2. In this case, the region R1 and the region R2 do not always match. In preparation for such a case, a region is defined by a union set when it is installed for a large number of subjects in a preliminary examination, and the region is set as an initial value of the functional localization agent.

  Normalized agent based on standard deviation: The output is normalized by constantly calculating the moving average and moving standard deviation for the output of an agent and dividing the current output result by subtracting the moving average by the moving standard deviation. This module

Normalization agent by pulse group: This module normalizes the output by constantly calculating the amplitude of the component corresponding to the pulse group and dividing by the value for the output of a certain agent.

Sigmoid conversion agent: A module that emphasizes and outputs a range to be separated by applying the sigmoid conversion process of Equation 3 to the output of a certain agent.

  AND agent: A module that receives two agent outputs and outputs the product.

  The maximum value agent is a module that receives a plurality of agent outputs and outputs the maximum value.

  Each agent module is a static filtering process executed sequentially, and each agent simply repeats the operation of creating an output for an input. Features for estimating the state of introspection activity for the first time by evaluating whether or not the result of this filtering process matches a specific activity, and performing synthesizer processing such as weighting according to the evaluation Can be converted to a value. The synthesizer part multiplies each of a total of M agents by an M-dimensional weighting value W = (w1, w2,..., WM) to obtain a total, and outputs the result to the output module. To do. A plurality of types of weighting are prepared in accordance with the number C of reflection tasks prepared as examples of commands.

  This “weighting value” is stored in the medium 605 and read out to the device 601 at the time of activation. In the initial state, the parameters learned for the average user are stored, but each time an operation to measure the coincidence between the activity and the agent behavior is performed in the measurement phase mode, the measurement results are accumulated and used. Thus, the weighting value can be improved in accordance with the characteristics of the individual, and the weighting parameter can be dynamically created according to the setting at that time.

  In addition, the weighting can be provided as a set of weighting W exceeding the number C of reflection tasks prepared as examples of commands. In such an example, if the learning calculation is performed by calculating the relevance ratio with the task using the output values of all agents, the learning calculation will not converge within a reasonable calculation time because there are too many variable dimensions. It is effective in some cases. In this case, a learning module with higher accuracy can be created by performing the above-described learning calculation for each predetermined subset and further adding the learning results. In order to synthesize this weighting value, it can be obtained by a machine learning technique called boosting.

  The above algorithm is executed as a program module on the information processing device 602 shown as the output unit in FIG. FIG. 9 shows a configuration example of a general-purpose computer that can execute such a program. Connected to CPU portion 901 and interface portion 902 for performing polishing processing, external storage device 903 (semiconductor media, optical media, magnetic media, etc.) for storing long-term data, image processing device 904, local area network (LAN) A device 905 and a main memory 906 are connected by a bus that mediates data, and an operating system 907, an execution program 908, a content resource 909 of display information, and the like are read from the external storage device 903 to the main memory 906 at the time of activation. A widely used technique can be used as it is for mounting a general input device 911 typified by a keyboard mouse or the like and a device 912 for accessing an external medium, such as operating a display monitor 910 from an image output device. . A technique for exchanging data with the information processing device 112 and the information processing apparatus 602 via the local area network 913 is also widely used. The information processing devices 112, 601, and 602 in FIG. 6 are all processing devices having such functions.

  FIG. 20 is a block diagram showing data contents for calling the filtering processing expressed in FIG. These data contents are stored in the external storage 903, and are read as a part of the execution program 908 on the main storage 906 at the time of execution. Reference numeral 2001 denotes a routine for managing the module connection state, and holds a tree structure as shown in FIG. Reference numeral 2002 denotes a routine for calling the function of the module program, and filtering calculation processing is performed by calling each API function of the module program according to a procedure defined by the routine. Reference numeral 2003 denotes the above-described module programs themselves. By setting the input relationship defined by the routine 2001 in this module program, the actual behavior is determined. 2004 is a routine for summing the outputs of the entire module, expressed as a synthesizer unit in the above description, and the output result of the entire evaluation function is derived by calculating a linear sum based on the weighting information stored in 2005. .

FIG. 21 is a flowchart representing an algorithm for calculating an evaluation value representing the current brain activity state using the data of FIG. A thread program operating in the background is called from 2101 every time data sampling is performed several times per second, and a series of analysis processing is performed. Reference numeral 2102 denotes an operation for calling an agent module and its input unit based on data recorded in the module 2001. At this time, if the called module is a primary agent, information from the associated sampler is used as its input, and if the called module is a secondary or tertiary agent, As input, output information of another associated agent is used. In response to these inputs, in step 2002, an algorithm defined for each agent is called to calculate the contents of the output (step 2103). This calculated content is stored in a buffer in step 2104. This stored value is used as input information to another agent, or is referred to when it is called from the synthesizer unit for final output calculation. The processing from 2102 to 2104 is repeatedly executed for the module elements registered in 2001 (step 2105).
The routine 2004 multiplies the output value of each agent module by the weighting value read from the data 2005 and outputs the sum (steps 2106 and 2107). This result is written into the latest buffer area, and the thread processing ends (step 2108). In each evaluation process described later, the latest data written in the buffer area is used.

In the operation process described in FIG. 21, the data 2005 read in step 2106 is rewritten based on the investigation result in the learning phase mode when executed in the real-time operation phase. This procedure and method will be described in the operation procedure of the learning phase mode described later.

The behavior of the device executed in the learning phase mode will be described subsequently using the real-time evaluation function created for each data sampling in the above filtering process. The content data stored in the information processing device 602 includes C examples of contents that present and explain introspection issues that can be executed by the user. When executed, it functions as an “instruction command”. This content module can be implemented by techniques such as screen display, audio display, and tactile display such as Braille. Examples of such tasks include processes such as singing a song in the head, making an image of moving a finger, calculating a subtraction, singing a song in the head, and performing a shiritori.
FIG. 11 shows a screen example of content that presents each task on the screen. It is an example of a display screen of a shiritori task 1101, a mental arithmetic task 1102, a moving image task 1103, and a song singing task 1104, respectively. In addition, by displaying an end screen such as 1105 at the task end time, the user can be prompted to end the task.

  FIG. 10 is a flowchart for explaining processing of a subroutine for realizing the calculation phase. First, the various initialization processing steps 1002 in FIG. 10 will be described. When a personal information medium is inserted, data is read and the device starts to drive. This media stores data when the user has operated this device in the past. The method for storing and using this data will be described later in the procedure. A laser is projected from the light irradiator (projector) 103 shown in FIG. At the same time, measurement by the receiver 105 is started, and the measuring device 112 writes information in a record and transmits it to the information processing device 601 at the same time. Thereafter, until the measurement is turned off in Step 1010, the NIRS sensor at each measurement position is periodically measured, and the oxidation and reduction and the total hemoglobin concentration are calculated from the light intensity passing through each wavelength, and information processing is performed. Continue to send to device 601. Further, the information processing apparatus 601 receives this information by using a thread programming technique widely known as a general function of the operation system (OS), and continues to update the information of the sampler 801.

  In the standby processing step 1003, a marker that blinks at regular intervals is displayed on the monitor 603. This marker indicates where to watch and indicates the timing of breathing. This process is repeated for a certain period of time until the fluctuation of the blood volume of the user 101 is stabilized. During this time, the user relaxes and waits so as not to perform excessive specific intellectual activities.

  Next, the problem presentation process 1004 will be described. One of the work task screens 1101 to 1104 is displayed on the screen of the monota 603. In the present embodiment, any of the above-described four types of task contents are selected in a random order and presented on the screen, and an instruction for prompting the user to execute the task is issued.

  In the pattern measuring step 1005, fluctuation data of the cerebral blood volume while the user is performing an activity in accordance with the instruction on the screen 603 is collected. An example of the collected data is given in FIG. Data obtained at the time of executing these tasks is sequentially sent to the agent function, and converted values are stored as time series information.

  When a certain time has elapsed, a screen 1105 for instructing the end of the task is presented (end processing step 1006). This is followed by a phase (rest processing step 1007) in which the subject is rested. During this time, the measurement is continued, and the fluctuation data of the cerebral blood volume that returns to the normal state after completing the execution of the task remains as data in the sampler. Further, the operations of the agent and the synthesizer continue to be performed, and the behavior is temporarily saved on the main memory as log data. Repeat the above steps a specified number of times.

  In this embodiment, an example was given in which all tasks were performed once by changing the order. However, a part of the task list may be executed at this time, or a specific task may be executed more than once. it can. However, it is preferable that a plurality of types of set tasks are executed.

  An evaluation of whether or not the agent module is moving in accordance with the mental function activity is performed in step 1009 of FIG. FIG. 12 is a block diagram expressing time series. The time series represents a graph in which the vertical axis indicates the output value of the agent and the horizontal time indicates the elapsed time. The time information about the timing presented to start / end the mental function activity is obtained via the network 913 in FIG.

  12, reference numeral 1203 in FIG. 12 indicates the time when the assignment is started (time t1), and reference numeral 1204 indicates the time when the assignment end instruction is presented (time t2). Yes. Reference numeral 1202 indicates a time t0 a certain time before the time t1, and reference numeral 1205 indicates a time t3 a certain time after the time t2.

  With respect to these times, the amounts of time delay required from the start / end of the activity until the change in blood volume is observed are denoted by Δ1 and Δ2, respectively. However, with respect to the values of Δ1 and Δ2, an appropriate range determined in advance from knowledge of biological measurement, such as from about 0 seconds to about 10 seconds, is used as the definition area. The period from t1 + Δ1 (time 1206) to t2 + Δ2 (time 1207), which is the time part in which the blood volume fluctuation can be observed, is defined as region A (1212), and the time parts (t0 to t1 + Δ1) 1211, (t2 + Δ2) where the blood volume fluctuation has returned. To t3) A region B is a collection of 1213.

  At this time, as a reference for determining the time delay values Δ1 and Δ2, in each of the regions A and B, the absolute value of the difference when the time-series average values (P1, P2) are calculated, | P1-P2 | = Δ1 and Δ2 satisfying the value that maximizes P0 are determined by searching from the range of the domain.

  As seen in FIG. 12, the value of the agent fluctuates even during the period of region A. The average and standard deviation of the fluctuation amount are calculated, and the value is set as σ1. Further, the standard deviation of this variation between the regions B is taken, and the value is set as σ2. For data i obtained by repeating a series of measurements, the value of P0 / (σ1 + σ2) is set as an evaluation value E (g, t, s) of the measurement data. Here, g is an agent number, t is a type of task executed introspectively, and s is an index value representing the sensor installation state in the setting. This calculation is performed for each agent and for the final output value of the synthesizer.

  After the calculation, the values of the data E and σ are stored in the medium 605. FIG. 13 is a diagram of a data structure showing an example of storing information in which the data is accumulated. An ordered number is given for each setting, and the value is described in a column 1301 in FIG. In addition, the setting time is stored in the column 1302. A plurality of tasks are executed for each setting, and a task type, a time, and an evaluation value of the result are stored in a column 1303, respectively. From these history data, the amount of dispersion for each task and the amount of dispersion for each setting can be obtained.

  Now, the setting s changes every time it is re-installed. T is a variable representing the type of task. When data measurement is repeated, E (I) is accumulated by the product of the number of measurements and the number of agents. After the measurement process 703 for each set of tasks is completed, the state evaluation value is estimated for the current setting s (process 704). For each agent, compare the results of this measurement with the results of past averages to calculate the confidence accuracy of the response in the current setting. An example of a technique for evaluating from a response in a specific setting using information on the result history for each task is shown below.

  In FIG. 7, the past history of each agent converted into parametric data by the operation 703 and its past operation is stored in the medium. As above, g is the agent number. t is a task number, and s is a value indicating a setting state. Information on the past history stored in the database is expressed as E (g, t, s, k). This indicates the k-th measurement result obtained under certain conditions [g, t, s].

  At this time, E (g, t, s, k) is considered to be a value that varies as a variable of t, g according to a certain probability distribution. For this E (g, t, s, k), the media 605 is defined as E0 (g, t), where E0 (g, t) is the average of all the histories of the measurement performed under the specific task t and agent g. Save to. The standard deviation is similarly stored as σ (g, t). Further, a value obtained by multiplying σ (g, t) by a constant is used as A (g, t), which is used as a reference value for the subsequent reaction of the agent to the task task.

  From this history, the sensitivity of the agent to the task is evaluated by the following procedure. A value obtained by dividing the average value Eo (g, t) by the A (g, t) electrode is defined as B (g, t). If this value is greater than 1, this agent g is judged to be an agent that exhibits a significantly positive response to the task t, and a positive weighting value W (g, t) = B (g, t) -1.0 is given. When B (g, t) is smaller than -1, negative weighting W (g, t) = B (g, t) +1.0 is performed. In other cases, W (g, t) = 0.0 is defined as an unstable state in which both positive and negative reactions are observed. These pieces of information are also stored in the medium 605.

  When data E1 (g, t, s, k) obtained by executing a certain task t under a certain setting s is obtained, the state of the setting s is evaluated using information stored in the medium 605. How to do will be described. E1 (g, t, s, k) is considered to be one sample of a probability distribution based on a normal distribution of mean E0 (t, g) and standard distribution σ (t, g) with respect to fluctuations in s, k. The null hypothesis is established and the verification process is performed.

  If this test result is significant, it is determined that the variation in the current setting s is significantly different from that in the normal setting environment. In this case, the task under the same condition t is repeated once more, and the number of k is increased and re-examination is performed. If this result is also significant, the screen 603 displays a message indicating that the behavior of agent g with respect to task t is strange, so that the position of the sensor associated with the sampler used as the input of agent g is displayed on the screen. Presented as illustrated above.

  If the above test result is not significant, it is determined that the change in the setting s of this time is within the allowable range as the behavior of the agent g, and E1 (g, t, s, k) for k Assuming that the average value is E2 (g, t), the value obtained by subtracting 1 from the absolute value of E2 (g, t) -A (t, g) as in the previous step is W1 (t, g in the current setting environment. ).

  After the above tests for all agents g and tasks t are completed, the data of W1 (t, g) is averaged and distributed for t. If the difference between the average of the W1 (t, g) values and the history values so far is significantly large, there is a high possibility that the agent state is not collected correctly.

  For this reason, display that the behavior of agent g is strange again, present the position of the sensor that the sampler used as the input of agent g is shown on the screen, and do not use this agent. Prompts you to choose whether to proceed with the following tasks or re-execute the measurement phase by re-setting. The average and variance of the data of W1 (t, g) are taken with respect to g. If the difference between the average of this value and the value of the history so far is significantly large, the task execution state is likely to be unstable. For this reason, a task with poor results is displayed, and the user 101 is reconfirmed with the user 101 whether there is a problem with the environmental state when this task is executed. The user 101 reconfirms and re-measures the state at the time of task execution, or avoids using the task as a command in the setting at that time, or weighting of the corresponding agent in this analysis process The value of the data 2005 read into the main memory is changed so as to decrease the ratio by a certain ratio.

  There are several advantages to using such an approach. The first advantage is that excessive demands on sensor contact can be avoided. Even if the location of the sensor is not in contact, if an important agent does not refer to it, the operation can be performed in the real-time operation phase even if it is left unattended. The second advantage is that past history information can be used for A (g, t), which requires a large number of trials for stable estimation. The third point is that it becomes possible to calculate a criterion for distinguishing between repetitive execution and a problem caused by task execution or a problem caused by agent reaction.

  From the result of step 703 (measurement phase mode), the calculation of the reaction state of the agent is performed, and the display of the sensor location presumed to be a problem in the acquisition of the target information and the current reaction state are efficiently performed. Display the combination of issues that can be acquired. An example of this display screen is shown in FIG. FIG. 14 shows a screen of the monitor 603. On the left side of the screen, a marker 1402 indicating a measurement point is arranged on a standard human model 1401 created as a three-dimensional model. On the right side of the screen, a list of agents 1403 related to physical information and information 1406 related to introspection tasks is displayed. Reliability accuracy 1404 and 1407 in this measurement, and standard reliability accuracy calculated from the history 1405 and 1408 are shown.

  As the reliability accuracy value, the value obtained by calculation can be used as it is, or the entertainment property can be improved by converting it to a value of 100 points. When the user selects one of the list elements, it is reflected in the color of the marker 1402 corresponding to the measurement point associated with the analysis of the agent. When a list element associated with the introspection task is selected, the amount of weight and the color density are reflected correspondingly. At this time, the reliability accuracy of the current sensor calculated in the previous step 703 is displayed as the blinking speed of the marker.

  The user confirms the result of the reliability accuracy related to this task, and in the case where the sensor is reinstalled and the measurement is started again, the recalculation console button 1408 is selected and the processing is started again from Step 702. In addition, when a problem that seems to satisfy the reliability accuracy that can be appropriately operated as a command is found, the selection determination console button 1409 is selected, and the process proceeds to the next real-time operation phase mode 706 (step 705).

  In the real-time operation phase mode (step 706), the user performs an introspective activity for changing the blood volume of the brain according to his / her intention, and the display is changed to reflect the result. By observing the results reflected by the user in this way, there is an effect of prompting conscious feedback in order to realize an appropriate operation. Hereinafter, the behavior of the device in the real-time operation phase mode will be described with reference to the block diagram shown in FIG.

  In the initialization processing step 1502, the variables required in the reflection update processing step described later are initialized. In addition, the past learning data stored in step 703 and step 704 is read. In a screen display process 1503, a selection screen similar to that displayed in the previous process 704 is displayed. In the method selection processing step 1504, the reliability accuracy of each introspection task calculated in the previous step 704 and displayed in step 705 is presented on the screen of the monitor 603. The user selects one of them. An operation can be performed by executing a reflection task as a command for transmitting a command in a reflection update process 1505 described later.

  The operation of the reflection update processing step 1505 is executed by a subroutine module. An example of mounting the module will be described with reference to the flowchart of FIG. An example of a screen image executed by this module is shown in FIG. On the screen 1700, a cursor 1701 operated by the user and a target position 1702 are presented. The target position 1702 is presented on the background screen, and the background screen scrolls downward at a constant speed for each update.

  In step 1602, initialization processing is performed. The position of the cursor 1701 is arranged at the center position of the screen.

  In step 1603, an initial screen is presented on the monitor 603.

  In step 1604, a short waiting process is performed to maintain the moving speed. This standby process is performed for several tens of milliseconds to several hundreds of milliseconds. While waiting for this screen update process, the measurement of cerebral blood volume data continues to be performed.

  In step 1605, the measurement result of the blood volume state at the latest time is applied to the signal processing system of the device 601, and the output result is calculated. In this process, when an operation such as a delay agent is used, information on the blood volume state in the past frame stored in the sampler 801 is also used for determination.

  In step 1606, information is updated. Moves the cursor position data to the left or right according to the output value above. Also move the background image down. Update the screen with the updated information.

In step 1607, the state is evaluated. If the cursor position data is included in the area of the target position 1702, the score is added.
As an end condition, it is confirmed that a specific score and a specified time satisfy a certain standard. If the termination condition is not satisfied, the above steps 1604 to 1607 are repeated. If the termination condition is met, proceed to step 1610.

  In step 1610, the end of the real-time investigation phase is notified on the screen of the monitor 603, the score is calculated, and the result is displayed. In addition, the result of this trial is temporarily stored in the variable area to be reflected in the learning data in the subsequent step 1508 (step 1610).

  This completes the subroutine 1505 of FIG. This is followed by a phase in which the subject is rested (rest processing step 1506). However, during this time, the measurement is continued, and the fluctuation data of the cerebral blood volume that returns to the normal state after the execution of the task is stored electronically in a log.

  The processing between the above steps 1503 to 1506 is repeated a specified number of times (for example, five times). When the specified number of times is reached, the investigation phase mode is terminated. The average score in the current operation phase 706 is stored in the medium 605.

  Compared to the average value and variance of the scores that have been compared so far in the media, the result (standard deviation) multiplied by a constant term is actually used as the correction value for this measurement. Is added to the value S (i) indicating the installation state of each agent used. Thus, the operation phase mode ends (step 1509).

  According to the first embodiment, when the localized brain function is measured by the optical brain function measuring device and the measurement signal is used as an input signal to the external device, selection and training for the task for the command command are performed. Depending on the learning depth of the person and the instability of the sensor, it can be used in combination according to different setting conditions every day.

  In the second embodiment, a subroutine module in the real-time operation phase mode is executed by an actual device in a computer. FIG. 18 is a configuration diagram illustrating an example of a device that is actually operated. In the second embodiment, instead of the cursor position data in the first embodiment, the position and angle of the arm 1801A are controlled for the shovel-shaped sorting device 1801. As in the first embodiment, the background moves, and in the second embodiment, the belt conveyor 1803 on which the article 1802 is placed moves in the direction of the thick arrow. Articles 1802 are periodically discharged from an article ejector 1805 onto a belt conveyor 1803. There is a hole 1804 on the belt conveyor 1803 so that the article can be dropped from the belt conveyor 1803 onto the first collector 1806 by moving the arm 1801A onto the hole 1804 at an appropriate timing. It has become. Articles that have not been dropped into the first collector 1806 by the sorting device 1801 fall into the second collector 1807.

  FIG. 19 is a block diagram of a general-purpose information processing device that can be used as a mounting device. The mounting device 1900 is another example of a general-purpose information processing device that is actually operated according to the present invention. In this general-purpose information processing apparatus, a plurality of lights 1901 and a sound source 1902 are installed. In step 1505, the current blood volume information is applied to the processing device 602, and an amount that matches the introspection task is calculated. In accordance with the calculated value, the amount of light, the volume of the sound source, and the pitch of the sound source are calculated. Changes. By performing this change constantly, it is possible to show a function as a simple musical instrument that is driven by a user's spontaneous operation.

  According to the second embodiment, when a localized brain function is measured by an optical brain function measuring device and the measurement signal is used as an input signal to a general-purpose information processing device, selection and training for a task for a command command Can be used in combination according to the situation of different settings every day according to the learning depth of the person and sensor instability.

It is a schematic diagram explaining the apparatus structural example of the measuring method in the external environment control apparatus by brain function measurement. It is a figure explaining the path | route of the light which propagates between the holder which fixes the optical fiber connected to the light irradiation device in FIG. 1, and the holder which fixes the optical fiber connected to the photodetector. It is explanatory drawing of the arrangement position of the optical fiber holder connected to the light irradiation device, and the arrangement position of the optical fiber holder connected to the photodetector. It is a figure explaining the example which specifies the fluctuation | variation location of a blood volume pattern with the some laser and sensor which are arrange | positioned at the grid | lattice form. It is a conceptual diagram of the method of isolate | separating multidimensional information linearly. It is a schematic diagram which shows Example 1 of the external control apparatus based on the biological light measurement by this invention. 3 is a flowchart illustrating an operation procedure according to the first embodiment. It is a system block diagram explaining the structure of the module which performs information processing. It is a block diagram of a general-purpose information processing device that can be used as an implementation. It is a flowchart of the subroutine which implement | achieves a calculation phase. It is explanatory drawing of the example of screen data of a subject. It is explanatory drawing of the example of data obtained by measurement of blood volume. It is explanatory drawing of the example of data which accumulate | stores past history information. It is explanatory drawing of the example of a screen which presents the starting state of an agent and prompts selection. It is a flowchart of the subroutine which implement | achieves an operation phase. It is a flowchart of the subroutine which implement | achieves a reflection update phase. It is explanatory drawing of the example of a screen displayed as an operation target in a reflection update phase. It is a figure explaining an example of the apparatus used as an operation target in a reflection update phase. It is a figure explaining another example of the apparatus used as an operation target in a reflection update phase. FIG. 9 is a block diagram showing data contents for calling a filtering process expressed in FIG. 8. FIG. 21 is a flowchart representing an algorithm for calculating an evaluation value representing the current brain activity state using the data of FIG. 20.

Explanation of symbols

101 Subject 102 Helmet for fixing sensor (probe, fiber holder)
103 Gold Infrared Laser Ejector 104 Laser Transfer Optical Fiber 105 Photodetector 106 Detection Optical Fiber 107 Optical Fiber Folder (Fixing Device)
108 Optical fiber tip device 109 Laser output control device 110 Laser output information transfer cable 111 Detector control device 112 Information processing device 113 for measurement unit Measurement cable 114 for measurement information transfer Screen monitor 115 for measurement unit Keyboard 116 for measurement unit Mouse 202 Scalp 203 Skull 204 Cerebrospinal fluid 205 Cerebral cortex 210 Shape of passage path of infrared laser 211 Cerebral cortex region whose metabolic rate has changed due to brain activity 301 Arrangement position of irradiation optical fiber 302 Arrangement position of detection optical fiber 303 Sampling point (measurement point)
401 Position 501 where brain activity is actively detected Vector position of data classified as state 1 502 Vector position of data classified as state-1 503 Hyperplane 601 separating state 1 and state-1 Information processing device 602 Information processing device 603 for output unit Information display screen monitor for subject 604 History information media reader 605 History information media 606 Information processing device control mouse 607 for output unit Module combination information reader 608 Module combination information Media 701 System start processing 702 in embodiment 1 Sensor helmet installation process 703 Measurement phase mode process 704 Agent reaction and recommended task display process 705 Reinstallation and task start selection process 706 Real-time operation phase mode process 707 Implementation End processing 801 in example 1 Sampler module 802 Primary agent module 803 Secondary agent module 804 Tertiary agent module 805 Pseudo agent module 806 Synthesizer module 901 Central information processing unit 902 Interface control unit 903 External storage device 904 Image output device 905 Local area network connection Device 906 Main storage device 907 Operating system 908 Execution program 909 Display information content 910 Display monitor 911 Input device 912 External media access device 913 Network 1001 Measurement phase start process 1002 Measurement phase initialization process 1003 Standby process process 1004 Problem task presentation Step 1005 Blood volume information measuring step 10 during execution of the task 6 Task task end processing 1007 User's break waiting step 1008 Loop end determination 1009 Analysis / save step 1010 Measurement phase end step 1101 Language recall task screen example 1102 Calculation task screen example 1103 Motorization task screen example 1104 Silent singing Example of Task Screen 1105 Example of Task End Declaration Screen 1201 Task Execution Period 1203 Task Start Timing 1204 Task End Timing 1206 Blood Volume Improvement Phase Start Timing 1207 Blood Volume Improvement Phase End Timing 1211 Preliminary Period 1212 Execution Period 1213 Post End Period 1301 Setting information field 1302 Task information field 1303 Agent evaluation information field 1304 Synthesizer output evaluation information field 1400 Agent reaction and recommendation task recommendation Display screen 1401 3D model of human head 1402 Marker 1403 corresponding to measurement point Agent characteristic information 1404 Evaluation value 1405 in this setting Average evaluation value 1406 in past history Task information 1407 Evaluation value in this setting 1407 Average evaluation value 1408 in past history Measurement selection button 1409 Selection decision button 1501 Operation phase mode start processing 1502 Initialization processing step 1503 Screen display step 1504 Method selection step 1505 Reflection update subroutine processing step 1506 Break time presentation step 1507 End condition determination step 1508 Analysis / save Step 1509 Operation phase mode end step 1601 Reflection update routine start processing 1602 Initialization processing step 1603 Initial screen presentation step 1604 Standby step 1605 Information analysis Step 1606 Screen and internal information update step 1607 Score addition step 1608 Termination condition determination step 1609 Score display step 1610 Reflection to learning data 1611 Reflection update routine termination processing
1700 Reflection update routine screen example 1701 Cursor 1702 Target point 1801 Excavator arm 1802 Article 1803 Belt conveyor 1804 Hole on belt conveyor 1805 Article ejector 1806 Article collector 1807 Article collector 1901 Lamp 1902 Sound source

Claims (3)

External environment control by brain function measurement that controls external devices according to changes in the blood volume state of multiple parts of the brain acquired by near infrared spectroscopy using a multi-channel fiber holder placed outside the head A device,
For each variation of the blood volume information obtained as time-series information from the fiber holder at each of the plurality of parts of the brain, for each biological function such as each region of the brain, pulse group, left-right difference of the brain, and its time difference A first means comprising a plurality of information processing modules for independently operating information to determine and extract a result;
Create a representative value obtained by summing the values output by each of the information processing modules constituting the first means according to the reflection task presented by the controlled device, and the representative value An external environment control device based on brain function measurement, comprising a second means for controlling an external device using a value obtained by performing a second filtering process. In claim 1,
A plurality of thirds that perform a task of changing the cerebral blood volume when the fiber holder is placed on the head and extract biological information such as each part of the brain, pulse group, left-right difference of the brain, and time difference thereof. With the means of
Evaluation / determination means for determining evaluation of detection accuracy of the biological information provided for each of the third means;
A storage unit for storing the evaluation / determination result by the evaluation / determination unit in an external medium for holding personal information;
Recalculating means for recalculating and storing the reliability accuracy for each of the third means from the average value and the variance value of a plurality of evaluation results according to the problems in the past history stored in the external medium. An external environment control device with characteristic brain function measurement. In claim 2,
Evaluating and comparing means for comparing the evaluation result of the response of the task detected at a time with a plurality of evaluation results and the average value / variance value of the evaluation results according to the task in the past history stored in the external media ,
Based on the comparison result of the evaluation comparison means, the installation state of the scalp of the head and the fiber holder and the measurement accuracy of the recalculation means are estimated, and optimal for the control of the external device according to the installation state of the fiber holder An external environment control device based on brain function measurement, comprising a monitor for displaying a screen for prompting selection or re-installation of the third means to be used with enhanced presentation and weighting of various problems.